EBDM

Question 1

[PP]
Systematic reviews and meta-analyses are generally agreed to provide the highest level of medical evidence.
Explain the main difference between systematic reviews and meta-analyses.

Answer 1

1. Systematic review
   = research method to select and review multiple studies
2. Meta-analyses
   = statistical analyses that combine results from different studies

Question 2

[PP]
Systematic reviews and meta-analyses are powerful research methods.
Define a systematic review and a meta-analysis.

Answer 2

1. Systematic review
   = gathers all available empirical research by using clearly defined, systematic methods (with pre-specified eligibility criteria)to obtain answers to a specific question (by providing an explicit and objective summary of evidence)
2. Meta-analysis
   = statistical process of analysing and combining results from several similar studies

Question 3

[PP]
The quality of systematic reviews and meta-analyses depend on the selected studies. Provide three reasons for this.

Answer 3

1. Eligibility criteria
   - The quality and reproducibility of each study included will influence the results of systematic reviews and meta-analyses
2. Internal validity criteria
   - Inclusion of individual studies with biases will influence systematic reviews and meta-analyses, e.g. publication bias
3. Internal validity criteria
   - Improper assessment of quality of evidence/results in the included studies will lead to misleading results in systematic reviews and meta-analyses

Question 4

[PP]
Give three reasons why systematic review and meta-analysis may not be appropriate if individual studies are too heterogeneous. (3 marks)

Answer 4

1. SR and MA are used to combine results from different studies for the same clinical question; however, if they are too heterogeneous, it may mean the studies were answering different clinical questions (e.g. different definition of disease, different intervention & outcome)
2. Different study designs (e.g. RCTs are negative, cohort studies are positive)
3. Different populations => confounding factors / effect modifiers (e.g. studies with Western population are positive, studies with Eastern population are negative)

Question 5

[PP]
A meta-analysis contains a forest plot that shows the adjusted estimates for the association of face mask use with viral infection causing COVID-19, SARS, or MERS.
Interpret the overall adjusted Odds ratio of 0.15 and its statistical significance.

Answer 5

Odds ratio of 0.15: when wearing a face mask, the chance of getting infection is 15% of not wearing a face mask.
Odds ratio < 1 => exposure is associated with a lower odds of outcome

Question 6

[PP]
What kind of bias can be reduced by blinding of the subjects?

Answer 6

Information bias

Question 7

[PP]
What kind of bias can be reduced by allocation concealment?

Answer 7

Selection bias

Question 8

[PP]
A recent, single-blind, randomised controlled trial was conducted to investigate the use of cognitive stimulation therapy (CST) to treat dementia. It reported that:
“At the post-test (26 weeks), there were no differences between the CST and TAU (treatment as usual) groups in the outcomes of cognition (mean difference [MD] = -0.55, 95% Cl -2.00 - 0.90; p=0.45) and self-reported QoL (quality of life) (MD = -0.02, 95% Cl -1.22 - 0.82; p=0.97) for people with dementia.”

Explain and interpret the 95% Cl for the outcome of cognition.

Answer 8

95% confidence interval (CI) means it is 95% confident that the true difference lies between -2.00 and 0.90. However, since the 95% CI crosses null value of 0, there is no statistically significant difference in cognition between CST and TAU groups
=> X find evidence to support the effectiveness of CST in improving cognitive function compared to TAU

Question 9

[PP]
An epidemiological study aims to correlate the average per capita consumption of red and processed meat in different countries across the world to their incidence of pancreatic cancer. What type of study design is it?
A. Case-control study
B. Cohort study
C. Cross-sectional study
D. Ecological study

Answer 9

D

Question 10

[PP]
Mrs. Chan, a 64-year-old clerk, recently went to a medical checkup. Her bone mineral density T-score was found to be -2.0 and therefore met the diagnostic criteria for osteopenia. As a result, Mrs. Chan became worried and asked whether she would have a higher risk of dying.
Formulate an answerable clinical question using the PICO format. (4 marks)

Answer 10

P: middle-aged Chinese women
I: with osteopenia
C: without osteopenia
O: mortality

Question 11

[PP]
Mrs. Chan, a 64-year-old clerk, recently went to a medical checkup. Her bone mineral density T-score was found to be -2.0 and therefore met the diagnostic criteria for osteopenia. As a result, Mrs. Chan became worried and asked whether she would have a higher risk of dying.
Name one patient-oriented outcome (apart from dying) that Mrs. Chan might also care about. (1 mark)

Answer 11

Quality of life
Disability
Pain
Costs of intervention

Question 12

[PP]
You are covering a busy medical ward overnight and you are informed by a nurse that by accident a patient has been given the wrong dose of insulin. Instead of administering 35 units of insulin the patient has received 55 units. The patient’s blood glucose is reading 1.0 mmol/L.
What kind of medical error is this? (1 mark)

Answer 12

Treatment error

Question 13

[PP]
You are covering a busy medical ward overnight and you are informed by a nurse that by accident a patient has been given the wrong dose of insulin. Instead of administering 35 units of insulin the patient has received 55 units. The patient’s blood glucose is reading 1.0 mmol/L.
What would you do? (4 marks)

Answer 13

IV dextrose to correct hypoglycaemia
Monitoring of condition (e.g. blood glucose level) until stable
Report the situation to senior for follow-up
Document the error in medical record
Explain and apologise to the patient

Question 14

[PP]
Name the BEST study design for answering questions about prognosis. (1 mark)

Answer 14

Prospective cohort study

Question 15

[PP]
Ms. Wong, a 70-year-old piano teacher, is suffering from chronic low back pain. Although she is troubled by the pain, she is most worried about whether she can continue to teach piano as it is her only source of income. She asks you about her prognosis.
Formulate an answerable clinical question on prognosis using the PICO format. (4 marks)

Answer 15

P: Elderly Chinese women
I: With chronic low back pain
C: Without chronic low back pain
O: Prognosis (e.g. functional ability including working, quality of life)

Question 16

[PP]

Observational studies show that high consumption of salted fish is positively associated with increased risk of nasopharyngeal cancer.
What type of observational study design using primary data would provide the highest level of
evidence of this association? (1 mark)

Answer 16

Prospective cohort study

Question 17

[PP]

Suggest two patient-oriented outcomes in observational studies relevant to patients with
nasopharyngeal cancer. (2 marks)

Answer 17

Mortality
Morbidity
Quality of life

Question 18

[PP]
Explain lead-time bias (1 mark) and how it may distort the evaluation of effectiveness of
cancer screening. (1 mark)

Answer 18

Lead-time: time between detection by screening and usual diagnosis
Lead-time bias: apparently improved survival time with no change in time from disease onset to death
-> make screening seemingly prolong survival

Question 19

[PP]
A meta-analysis of 9 clinical trials compared outcomes of bariatric surgery and non surgical treatment among patients with obesity and type II diabetes. Patients that received surgery had a higher remission rate of diabetes compared to non-surgically treated patients (relative risk 20.1 with a confidence interval of 2.1 to 109.4), but the heterogeneity of the combined study results was high (I2 = 97%). Reading the forest plot, you realise that studies that used adjusted gastric band had a lower relative risk compared to other bariatric surgery techniques. What is recommended to perform to deal with the high heterogeneity in this meta-analysis? A. Egger test B. Fixed-effect model C. Funnel plot D. Subgroup analysis

Answer 19

D

Egger test: assess publication bias, test of asymmetry of funnel plots predicting discordance of results in meta-analysis

Fixed-effect model: statistical regression model, estimate effect of intrinsic characteristics of a data set (e.g. genetic factors) which are not directly observable, DV change but IV no change

Funnel plot: scatter plot, effect estimates on horizontal axis, study size on vertical axis, assess publication bias

Question 20

[PP]
A new drug to treat acute migraine attacks in adults proved complete pain relief among 35% of the patients compared to 10% among patients receiving placebo. What is the number needed to treat with the new drug to prevent a migraine attack? A. 2 B. 4 C. 25 D. 40

Answer 20

B

NNT = 1/ARR
ARR = 0.35 - 0.1 = 0.25

Question 21

[PP]
An epidemiological study aims to correlate the average per capita consumption of red and processed meat in different countries across the world to their incidence of pancreatic cancer. What type of study design is it? A. Case-control study B. Cohort study C. Cross-sectional study D. Ecological study

Answer 21

D

Unit of observation: individual VS group

Cross-sectional analysis: generate prevalence, X incidence

Question 22

[PP]
Prevention includes a wide range of activities aiming at reducing risks or threats to health. Prevention activities are typically categorised in three levels. What is the aim of tertiary prevention? A. To reduce complications of disease B. To reduce incidence of disease C. To reduce overdiagnosis D. To reduce prevalence of disease

Answer 22

A

Primary: reduce incidence
Secondary: prevent progression, reduce overdiagnosis

Reduce prevalence: due to cure or death, can occur at secondary or tertiary levels

Question 23

[PP]
In a clinical trial, 60 patients were assigned to the treatment group, 120 patients were assigned to the control group. By the end of the trial, 6 patients died among the 60 patients in the treatment group, and 24 died among the 120 patients in the control group. What is the Number Needed to Treat (NNT) to prevent one death? A. 1 B. 2 C. 10 D. 20

Answer 23

C

Absolute risk reduction = 0.1

Question 24

[PP]
Multiple diagnostic tests are often required in clinical settings to diagnose a disease or a condition. Multiple tests can be performed in parallel or in series. In parallel testing, two or more tests are performed at the same time. When the tests results are negative, parallel testing helps to ‘rule out’ a disease aided by which of the following? A. Decrease in negative predictive value B. Increase in positive predictive value C. Increase in the sensitivity D. Increase in the specificity

Answer 24

C

Parallel testing: increase SN and NPV, but may increase false positives

Serial testing: increase SP and PPV, but may increase false negatives

Question 25

[PP] Dr. Leung wants to investigate the prognosis of obesity people who is diagnosed with rheumatoid arthritis. What is the BEST individual study design to determine whether breastfeeding is a prognostic factor? A. Case-control study B. Prospective cohort study C. Randomised-controlled trial D. Retrospective cohort study

Answer 25

B

Question 26

[PP] Patient harm is a leading cause of global morbidity and mortality. Moreover, unsafe care leads to increased healthcare costs and loss of trust in health care systems. Following the approach of the Institute of Medicine of the USA and the report "To Err Is Human", what are the main causes of medical errors? A. Faulty systems, processes, and conditions that lead people to make mistakes or fail to prevent them. B. Individual recklessness and actions of particular groups of healthcare workers that harm the patients intentionally. C. Lack of post marketing surveillance of drugs under real-world conditions, since drug adverse reactions are not monitored after roval. D. There are no main causes of medical errors, they are unpredictable and unpreventable.

Answer 26

A

Question 27

[PP] A phase III placebo-controlled clinical trial randomly assigned participants to receive either two intra-muscular injections of a vaccine or a placebo to prevent coronavirus disease. Trial staff that performed assessments and interacted with participants were unaware of the intervention that participants received. Concerning staff blinding, what type of bias did researchers intend to prevent? A. Concealment bias B. Information bias C. Lead-time bias D. Selection bias

Answer 27

B

Question 28

[PP] A 50-year-old male patient comes to his primary care doctor to discuss colorectal cancer screening. He wants to know how much a positive result for faecal occult blood test would increase the chances of having colorectal cancer. Which measure is the MOST ADEQUATE to answer his question? A. Negative predictive value B. Positive likelihood ratio C. Positive predictive value D. Sensitivity

Answer 28

B

Question 29

[PP] SYSTEMATIC REVIEW AND META-ANALYSIS A. Fixed effect model B. Forest plot C. Funnel plot D. Jadad score E. Meta-analysis F. Newcastle-Ottawa quality assessment scale G. PRISMA checklist H. Random effects model I. Sensitivity analysis J. Systematic review Twenty randomised-controlled trials (RCT) were included in the meta-analysis, which concluded that the exposure is significantly associated with the primary outcome. However, it is not known whether the result will remain significant if all RCT without allocation concealment are removed from the analysis.

Answer 29

I Test whether the conclusion is sensitive to some parameters of the original analysis being modified If insensitive -> robust

Question 30

[PP] SYSTEMATIC REVIEW AND META-ANALYSIS A. Fixed effect model B. Forest plot C. Funnel plot D. Jadad score E. Meta-analysis F. Newcastle-Ottawa quality assessment scale G. PRISMA checklist H. Random effects model I. Sensitivity analysis J. Systematic review The meta-analysis included 28 individual studies. Although the researchers established explicit inclusion and exclusion criteria, they were worried that publication bias might still be present.

Answer 30

C

Question 31

[PP] SYSTEMATIC REVIEW AND META-ANALYSIS A. Fixed effect model B. Forest plot C. Funnel plot D. Jadad score E. Meta-analysis F. Newcastle-Ottawa quality assessment scale G. PRISMA checklist H. Random effects model I. Sensitivity analysis J. Systematic review Ten cohort studies and ten case-control studies were included in the meta-analysis. The two types of study seem to have very heterogeneous results, suggesting different underlying effects.

Answer 31

H Allow for different underlying effects between studies other than sampling variation

Question 32

[PP] A low-iodine diet increases radioactive iodine uptake and is routinely recommended for thyroid cancers. However, its role in Graves' diseases is less clear. What type of journal article would provide the best evidence to help you make the clinical decision to start a patient on a low-iodine diet who is receiving radioactive iodine for Graves disease? A. A narrative review authored by international thyroid disease experts B. An original paper reporting a multi-centre, randomised controlled trial C. A systematic review and meta-analysis of cohort studies D. A systematic review and meta-analysis of randomised trials

Answer 32

D

Question 33

[PP] Dr. Chan is conducting a randomised-controlled trial to investigate the efficacy of a new therapy for lymphoma. In order to avoid investigators cherry-picking healthier patients to the desired treatment group and introduce selection bias, what technique shall be employed? A. Allocation concealment B. Blinding of the assessor C. Double blinding D. Intention-to-treat analysis

Answer 33

A

Question 34

[PP] At a cut-off of 20 ng/mL, the sensitivity of alpha-fetoprotein (AFP) for screening of hepatocellular carcinoma (HCC) is approximately 40%-60% and its specificity is approximately 80%-90%. Increasing the cut-off value of AFP would be expected to have what effect on the test's validity? A. Decreased negative predictive value B. Increased positive predictive value C. Increased sensitivity D. Increased specificity

Answer 34

D

Question 35

[PP] ELEMENTS OF CLINICAL TRIALS A. Allocation concealment B. Blinding C. Equipoise D. Equivalence E. Explanatory trial F. Intention-to-treat analysis G. Per-protocol analysis H. Pragmatic trial I. Randomisation J. Sensitivity analysis A clinical trial was designed to demonstrate that a new orally administered iron chelator is therapeutically similar to an existing subcutaneous iron chelator.

Answer 35

D Equivalence trials: no better and no worse E.g. a new drug with similar effectiveness is safer, cheaper or easier to administer

Question 36

[PP] ELEMENTS OF CLINICAL TRIALS A. Allocation concealment B. Blinding C. Equipoise D. Equivalence E. Explanatory trial F. Intention-to-treat analysis G. Per-protocol analysis H. Pragmatic trial I. Randomisation J. Sensitivity analysis In a clinical trial, the investigators were not aware of each participant's group assignment before enrolling them into the study.

Answer 36

A

Question 37

[PP] ELEMENTS OF CLINICAL TRIALS A. Allocation concealment B. Blinding C. Equipoise D. Equivalence E. Explanatory trial F. Intention-to-treat analysis G. Per-protocol analysis H. Pragmatic trial I. Randomisation J. Sensitivity analysis In a clinical trial, the treatment group analysis is included of those patients who completed the treatment originally allocated.

Answer 37

G Analysis including participants only if they received the intended intervention in accordance with the study protocol = Per protocol analysis Intention-to-treat analysis = analyzing study outcome in participants no matter whether they completed the treatment originally allocated - preserves the randomisation

Question 38

[PP] ELEMENTS OF CLINICAL TRIALS A. Allocation concealment B. Blinding C. Equipoise D. Equivalence E. Explanatory trial F. Intention-to-treat analysis G. Per-protocol analysis H. Pragmatic trial I. Randomisation J. Sensitivity analysis A clinical trial involving 100 female patients with end-stage chronic renal disease found that a new drug was more beneficial compared to an existing drug in increasing the patients' haemoglobin level.

Answer 38

E Explanatory (efficacy) trial - highly selective and homogeneous patients Pragmatic (effectiveness) trial - routine health care setting - more diverse patients

Question 39

[PP] In conducting a systematic review, usually at least two independent reviewers are needed to conduct literature search and data extraction. Which type of bias could this minimise? A. Citation bias B. Information bias C. Publication bias D. Recall bias E. Selection bias

Answer 39

E

Question 40

[PP] The age-adjusted incidence and mortality of pancreatic cancer have risen in the past ten years. You undertake a literature review for suitable screening programmes of pancreatic cancer. Observational studies show inconsistent results as they are susceptible to bias and confounding factors. What bias may arise from comparing survival time with and without screening in observational studies? A. Information bias B. Lead-time bias C. Length-time bias D. Selection bias E. Survivor bias

Answer 40

B To avoid: compare mortality rather than survival rates

Question 41

[PP] A group of researchers is planning to study the effectiveness of a new analgesic drug in comparison to paracetamol, and is considering the number of subjects needed for running a randomised controlled trial. Which type of error or bias is reduced when the sample size of a study increases? A. Confounding B. Information bias C. Selection bias D. Type I error E. Type II error

Answer 41

E Likelihood of capturing the effect of an association (if one exists) increases -> reduces the error of failing to reject a false null hypothesis (i.e. a type II error) Null hypothesis: no difference between the groups being compared, i.e., no difference, no association, or no effect Type 1 error: falsely rejecting a null hypothesis when it is true (false positive result) Type 2 error: falsely failing to reject a null hypothesis when it is false (false negative result)

Question 42

[PP] A randomised study sample in a randomised-controlled trial (RCT) is not being representative of the source population it is drawn from, rendering a limited generalisability to the study's results. What is the prime reason for the limited generalisability? A. High cost B. Long follow-up period C. Loose inclusion criteria D. Sampling bias

Answer 42

D

Question 43

[PP] Climate change could have far-reaching and lasting consequences for society. It is important to use an appropriate technique to assess the future benefits and costs of interventions to reduce greenhouse gases. What technique makes a fair comparison of interventions if costs and outcomes occur at different times? A. Discounting B. Incremental cost-effectiveness ratio C. Marginal analysis D. Negating

Answer 43

A

Question 44

[PP, formative] The prevalence of a certain musculoskeletal disease is 25%. The available diagnostic test has a sensitivity of 80% and specificity of 90%. What are the post-test odds of having been diagnosed with the musculoskeletal disease? A. 0.33 B. 0.75 C. 2.64 D. 8.00

Answer 44

C